alternatives to text

Concerning   
                       WAI Accessibility Guidelines:
                           Browser User Interface
          http://www.w3.org/WAI/UA/WD-WAI-UA-BROWSER-19980603

in section
       
  4.2 Alternative Representations of Images

Comment 1:

where it says..  
    1. [PRIORITY 1]
       Allow the user to turn off the display of images inserted by
       HTML's IMG element ...

and then
    2. [PRIORITY 1]
       Allow the user to turn off the display of images inserted by
       HTML's OBJECT element 

It is not clear to me that two separate controls are necessary to meet
accessibility requirements (or desirable).  If the function of the
user mode selection is to block images, then the element type 
bearing the image is not a factor in the decision and one guideline
here would be clearer than two.

Comment 2:

It may be helpful for the guidelines to clarify the processing of
links with image content in a mode where images are not
automatically displayed but displayed on individual selection by
the user.  In such a mode the user should have the capability to
follow the link without having to display the image first.  If
the behavior of browsers were more consistent in this regard, the
conformance of authors to guidelines for markup attributes would
be better.

Comment 3:

in
    2. [PRIORITY 1]
where it says
                          ... The innermost text of the OBJECT is
       considered its alternative text. 

I believe that this is not entirely consistent with the HTML
specification, and that following the HTML specification in this
instance is sufficient; we don't need to bend the rules here. If
there are nested OBJECTs, each content may include fairly general
HTML hypertext an not just OBJECT elements.  Each OBJECT for
which the data are not presented contributes its entire content
and not just OBJECTs appearing in its content to the effective
document to be presented.  If there are three forms of image
cited as data in three nested OBJECTs and note meet the current
(system and user) requirements for display, then the full
remaining content of all levels is to be displayed, I believe.
Any text appearing in the content of outer OBJECTs should not be
skipped in the construction of what to present to the user.

Note that under the fallback rules for OBJECT in HTML 4, the 
data for an outer object could be a video with content comprising
text and an audio object.  Even if this second object had a
content of pure text, that innermost text would not necessarily
be presented to the user if the audio in the inner OBJECT meets
system and user requirements to be presented.

Comment 4:

In the final three rules in this section, viz:

    3. [PRIORITY 1]
       When an IMG element has a value for the "longdesc" attribute and
       the user has turned off the display of images, render a D-link
       inline to give access to the long description. Provide keyboard
       access to locate and select the long description (in addition to
       pointer access for able-bodied users).
    4. [PRIORITY 3]
       When an IMG element has a value for the "longdesc" attribute and
       the author has already defined a D-link for the image, the
       "longdesc" D-link should be suppressed. Therefore if an IMG
       element has both a value for "longdesc" and a hard-coded D-link
       only one D-link should be presented to the user.
    5. [PRIORITY 3]
       A user selectable option is available to turn off the display of
       hard-coded D-link in a document. The D-link should assign the
       valus of "dlink" using the REL attribute of the ANCHOR element.
       All links using having this attribute would be hidden by the
       browser.
       
I am not entirely optimistic about the author adoption of this dlink
REL value.  I think we may get better results if we ask the browser
to compare the URL in the LONGDESC with the value of href atttributes
in the vicinity.

In other words, consider something on the order of:

a. The browser must provide a mode where any description which is
the value of a LONGDESC attribute in an IMG element is available
as a navigation destination via a link in the vicinity of the
image.  In particular, the browser must synthesize some sort of a
link if there is not a link to that resource in the vicinity of the
image, and the browser may omit the synthetic link if the browser
detects that there is a link to the same resource in the vicinity
of the image.  The vicinity of the image is implementation
defined.

b. The browser may provide modes under which there is more selective
display of a synthetic link to resources identified in LONGDESC.

I am not totally optimistic about the effects on d-link usage by
authors by making the suppression of links with REL="dlink" an
option in some browsers.

If we go forward with an option to suppress links marked with a
"dlink" attribute, I would suggest that CLASS="dlink" is more
appropriate than REL="dlink" in this case.  D-link is a
subcategory of link and the A element with a live HREF value is a
member of the "link" category.  But the sense of REL is that it
is a relationship between the current document and the referenced
resource, and that is not "dlink."  The dominant semantic
relationship is between the referenced resource and _the image_
and we didn't get tools to annotate that outside of providing a
LONGDESC on the IMG.

Al

Received on Friday, 5 June 1998 15:41:28 UTC